Ejemplo n.º 1
0
 def test_extract_page(self):
     url = 'http://esf.sh.fang.com/house/g22-j280-k2100-l3010-kw%c3%fb%b6%bc%d0%c2%b3%c7'
     _, html = spider.get(url)
     # test for extract
     total, next_link, houses = extract_soufan.extract_page(html)
     self.assertIsNotNone(total)
     self.assertIsNotNone(houses)
     self.assertTrue(int(total) > 0)
     self.assertTrue(len(houses) > 0)
     for house in houses:
         #self.assertEquals('名都新城', house.name)
         self.assertIsNotNone(house.name)
         self.assertIsNotNone(house.village_name)
         self.assertIsNotNone(house.storey)
         self.assertTrue(house.orientation != -1)
         self.assertTrue(house.age != -1)
         self.assertTrue(house.age > 1990)
         self.assertTrue(house.building != -1)
         self.assertIsNotNone(house.address)
         self.assertIsNotNone(house.agent)
         self.assertIsNotNone(house.policy)
         self.assertTrue(house.total_price != -1)
         self.assertTrue(house.total_price > 1000000)
         self.assertTrue(house.centiare_price != -1)
         self.assertTrue(house.centiare_price > 1000)
     self.assertIsNotNone(next_link)
Ejemplo n.º 2
0
 def test_get_shanghai_phoenix(self):
     url = 'http://esf.sh.fang.com/house/g22-kw%ba%c3%ca%c0%b7%ef%bb%cb%b3%c7'
     module = 'soufun'
     root_link = 'http://esf.sh.fang.com'
     house_page = spider.get(url, module, root_link)
     self.assertIsNotNone(house_page)
     self.assertTrue(len(house_page.houses) > 0)
Ejemplo n.º 3
0
 def test_extract_page(self):
     url = 'http://esf.sh.fang.com/house/g22-j280-k2100-l3010-kw%c3%fb%b6%bc%d0%c2%b3%c7'
     _, html = spider.get(url)
     # test for extract
     total, next_link, houses = extract_soufan.extract_page(html)
     self.assertIsNotNone(total)
     self.assertIsNotNone(houses)
     self.assertTrue(int(total) > 0)
     self.assertTrue(len(houses) > 0)
     for house in houses:
         #self.assertEquals('名都新城', house.name)
         self.assertIsNotNone(house.name)
         self.assertIsNotNone(house.village_name)
         self.assertIsNotNone(house.storey)
         self.assertTrue(house.orientation != -1)
         self.assertTrue(house.age != -1)
         self.assertTrue(house.age > 1990)
         self.assertTrue(house.building != -1)
         self.assertIsNotNone(house.address)
         self.assertIsNotNone(house.agent)
         self.assertIsNotNone(house.policy)
         self.assertTrue(house.total_price != -1)
         self.assertTrue(house.total_price > 1000000)
         self.assertTrue(house.centiare_price != -1)
         self.assertTrue(house.centiare_price > 1000)
     self.assertIsNotNone(next_link)
Ejemplo n.º 4
0
 def test_extract(self):
     url = 'http://esf.sh.fang.com/house/g22-j280-k2100-l3010-kw%c3%fb%b6%bc%d0%c2%b3%c7'
     _, html = spider.get(url)
     # test for extract
     house_page = extract_soufan.extract(html)
     self.assertIsNotNone(house_page)
     self.assertTrue(house_page.total > 0)
     self.assertTrue(house_page.next_link > 0)
     self.assertTrue(len(house_page.houses) > 0)
Ejemplo n.º 5
0
 def test_extract(self):
     url = 'http://esf.sh.fang.com/house/g22-j280-k2100-l3010-kw%c3%fb%b6%bc%d0%c2%b3%c7'
     _, html = spider.get(url)
     # test for extract
     house_page = extract_soufan.extract(html)
     self.assertIsNotNone(house_page)
     self.assertTrue(house_page.total > 0)
     self.assertTrue(house_page.next_link > 0)
     self.assertTrue(len(house_page.houses) > 0)
Ejemplo n.º 6
0
 def test_get(self):
     url = 'http://esf.sh.fang.com/house/g22-j280-k2100-l3010-kw%c3%fb%b6%bc%d0%c2%b3%c7'
     module = 'soufun'
     root_link = 'http://esf.sh.fang.com'
     result = spider.get(url, module, root_link)
     self.assertIsNotNone(result)
Ejemplo n.º 7
0
 def update_if_newer_version(self):
     latest_version, latest_version_url,_ = \
                             spider.get(self.is_new_version_url).split('\n')
     if self.__is_newer_version(latest_version):
         self.up_to_date = False
         self.update_to_latest_version_given_url(latest_version_url)
Ejemplo n.º 8
0
def main():
    ms = DBMssql.MSSQL(host="localhost", user="******", pwd="qq123456", db="scxt")
    courseOfImooc.createTable(ms)
    list = spider.get()
    courseOfImooc.store(ms, list)
    draw(ms)